That's hardly comparable. Whether overpriced or not, the excuse for expensive drugs is that it pays for the development. I can assure you that the development of an ink cartridge cost a tiny fraction of the manufacturing price.
Now maybe if big Pharma built free hospitals for everybody and then relied on the prices of the drugs used by those hospitals it would be similar to what the printer companies are doing.
You can take the Heathrow Express, which takes about 15 minutes and costs a lot (15 pounds?). It drops you at Paddington station, which is the northwest corner of the interesting area and all the tube lines go there.
If you have more time a cheaper option that really works is to take the Picadilly line in. This will cost less than 4 pounds (get an Oyster card even if you are there for only one day), and you are already on the tube so you can get to whatever station you really need for no extra cost. However travel time is like 50 minutes. Most of it is above ground so you can see something.
You can still get plenty of news for free on the Internet!
Did you know that George Bush parachuted out of the airplanes just before they hit the WTC? I would not know that except for reading the free news. Also did you know that Al Gore is using global warming as a smokescreen to hide the thermal exhaust from his secret base under the ice cap from which he will enslave the world?
By hard disks they meant "storage with a moving part", not mass-storage. Flash and other solid state memory is allowed, especially removable ones. I really don't see how they could prevent external USB hard disks however, without also disabling most USB flash disks.
My GF uses aol mail to read all her email. On several machines (OS/X) I have correctly set up Mac mail and even Thunderbird to read/write her AOL email, which worked great, and she still uses the web page to do it.
This really convinces me that there will be zero problems with putting everything into the browser. AOL mail is far worse than gmail.
According the the etymology I read the BDSM practice is named after the character in Pulp Fiction, who was named after the common use of the word "gimp".
In any case I think it is silly to say this when the primary complaint is that for a lot of people "gimp" means "cripple" or "disabled person" or whatever the term is. Bringing up a BSDM term that may very well be more obscure than "GNU Image ManiPulator" does not advance the argument.
Can other programs (such as the image viewer such as EOG that you get if you double-click an image) print reasonably correctly? If so then you have seen exactly a problem with color management: when it screws up it screws up good.
If other programs also seem to mess up in a similar way I would say the printer driver is screwed.
The question I have is why everybody complains about this when it is THE EXACT SAME THING in Photoshop? What the hell, is the fact that it is Photoshop somehow make it ok?
You seem to be posting this a lot. Actually "gimp" is slang for a disabled person (I think it may imply physical rather than mental disability). The "sexual perversion" is from a name of a bondage slave character in the movie Plup Fiction, the name obviously being chosen from the common meaning of "gimp", not the term for the bondage itself.
If you are going to question the name of GIMP then at least get your facts right.
You hold down shift and click to draw a straight line.
Please explain how you draw a straight line in Photoshop. I searched all over, because I have heard this complaint, but the only thing I found was "hold down shift and click". It is EXACTLY THE SAME (except Gimp has a overlaid line preview which makes it a lot better). I may be an idiot, but I tried all the polygonal tools and everything, nothing drew a straight line.
I am certainly not a professional user in any way, but have used Gimp (the one with Ubuntu 9.04) and Photoshop (most recent on OS/X) about equally.
The windows seem to be pretty much the same. Gimp creates a single one on the left, I have managed to get more than one but hate it, and it was frustrating to mess with it and try to figure out how to get it back into a single stacked window. I think they fixed things considerably by moving the menu bar from the tool window to the "main" window. Photoshop seems to create two big floating windows, one on the left and the other on the right.
I like Gimp's panel that is the current tool options. Photoshop seems to hide the options for the tools in different places depending on the tools. I really dislike the Photoshop slide-out windows on those things on the right and wish they would just put them in tabs in the floating windows like some other things.
I like the fact that you can differentiate the mouse and pen in Gimp and get three tools. Photoshop seems to think the mouse and pen tip are the same tool. And in Gimp if I use the eraser end to pick brushes, I change the eraser's brush, in Photoshop it changes the pen's brush (the only way to change the eraser seems to be to pick eraser with the pen tip, change it, then put the pen tip back). For that I give Gimp a big win in usability. I also think Gimp is enormously better at zooming with the mouse wheel, Photoshop has some very squirrly results when you do this that I really cannot explain on a modern program...
Photoshop seems to make better smoother brush strokes. Gimp had a horrible problem that antialiasing disappeared if you painted when zoomed out more than 1:1 but they seemed to have fixed that, however that sort of bug is pretty close to inexcusable, sorry. Also all brushes other than the default solid ones are useless hacks in Gimp, but appear to produce real painterly effects in Photoshop.
A huge obvious technical benefit of Photoshop is layer grouping (really a tree of compositing operations rather than just a line).
Both programs suffer from an enormous number of mystery components. What do the "chains" mean in the layers? I have clicked them on/off in both and can't see what they do. Selecting regions, cutting and pasting, and somehow getting the pasted thing to "stick" is a complete mystery.
That's a silly complaint. If you typed some text, it would have changed to the selected font as you picked it from the menu. Photoshop has an option to set the menu this way as well. It is useful because sometimes the font names are unreadable when written in the font itself, and I certainly have set it this way.
(the fact that Linux lists a whole lot of useless fonts such as a zillion ones that are really just some foreign subset of letters and don't change the main font is a bug. However I was shocked when I tried a recent version of Windows and saw identical enormous lists of useless fonts, I guess they think that was some killer feature of Linux that they wanted to copy???)
I'm reasonably certain even the cheapest photo editor that claims "red eye removal" does it by manipulating the red channel differently than the other channels. Most common solution is to set r = min(r,max(g,b)) in the locations you paint. Slightly more complex is to multiply the result so that the overall brightness remains the same, you don't even have to weight the channels differently. This is rather harmless even if you apply it to the entire image, so simply doing this to an eye-sized blob of the picture, even with no antialiasing of the edges, will work.
In any case I think even the goofballs who write the crap like the "free photo software included with your camera!" can do this correctly and would be suprised if they got it wrong.
Though dweebs here like to throw out buzzwords like CMYK and > 8 bits, the most obvious missing thing is that you cannot "group" the layers so that the compositing operation is done between them and then the result is overlayed. For instance you cannot non-destructively colorize a lineart layer and put it on top of a background, something that Photoshop makes easy.
More than 3 channels (CMYK is one minor use of that) would be nice. In professional special effects graphics these are used mostly for mattes and effects channels and information such as the normals of the surfaces. Use for the printing "black" is a minor insignificant detail compared to these other things.
Having worked with professional graphics quite a bit I have to say that "color management" is 95% bullshit. It is not possible to make a reflective printout the "same" as a light-emitting screen, anybody claiming this is lying.
Photo manipulation and painting is helped considerably by not losing information on display, this means that on current 8-bit images and 8-bit displays, any method other than 1:1 mapping of the image values to the display is WRONG, and thus most "color management" is in fact harmful (dithering and error diffusion can resolve this problem some, but nobody is doing it because users don't like the slightly-visible patterns, 10-bit displays may help here).
If you really want to manage actual light data, the most important step is to change the internal representation to a "linear" format where the emitted energy is proportional to the stored number, but the "color management" people refuse to do it because it would make "color management" (ie changing the primaries) into a trivial matrix transform and put them out of business. Also it is not practical in any integer-based storage format.
I very much hope they forget completely about any integers > 8 bits. If you are going to use 16 bits then use ILM/Nvidia "half" floating-point format. Stop living in the previous century and pretending something Photoshop did then is actually modern...
What's the difference between putting a picture into your "vacation 2009" and tagging it with "vacation" and "2009"?
The difference is that if you then decide to tag half of them with "summer" then that half are no longer in the "vacation 2009" directory.
Symbolic links can make this somewhat work, but as soon as you have a number of tags that form intersecting sets the number of directory entries grows to m*2^n (where m is the number of files and n is the number of different tags). This is far larger than the m*n space needed by tags even with a stupid implementation.
I do however believe a working system can be built where POSIX file api still can be used. A tag is a directory, and a whole tree of them is just a directory listing the files in that. So./vacation/2009/ and./2009/vacation/ would both have the same files: all the ones tagged with those../vacation/2009/summer would have files tagged with all three.
This is exactly what should be done. But not with "metadata" as people are saying. We need "metadata" embedded directly into the files. Then it is preserved when the file is copied, archieve, or stored on a system that has a different concept of metadata.
The normal implementation of "metadata" would then be used to cache the metadata read out of the file. For the best example of how this works, imagine the "thumbnail images" that some systems are storing. These are stored in the file system in some way, but are completely figured out from the original file. If the thumbnail metadata disappeared it would get repliated again later. What I would like to see is all metadata is produced from the file contents in this same way.
The way to do it is to make up some rules for how to find it, based on the file types. This does require a program that understands a lot of file formats, but that would be an actual executable program, not a "library" and thus easy to expand and for any program to call. It would for instance understand how to extract jpeg comments. For text files it would know how to recognize comments and then rules could be made up for patterns in the comments to indicate the metadata.
I could see this eventually replacing all "metadata" including long-standing things such as the name, modification date, and even the protection bits of the file (the actual protection would be the intersection of some guarded protection and what the file says, but a program with sufficient privledges could update the real protection to match the file data).
When offering something up like this is the company expected to just put up a big banner at the top saying, "HEY, WE ARE CHARGING YOU FOR SOMETHING IF YOU CLICK YES!" before even trying to sell the person on the product?
YES! That is an excellent description of exactly what should be required.
However the upscaling is usually used to show the Windows startup screen, so an improved upscaler (or just making it center a 1:1 image in the middle of the screen) would make a lot of buyers go "hey this screen looks a lot better" when they turn their computer on.
So I agree it is a mystery why they go for such fuzzy awful upscaling.
A lot of LCD screens have an option to center the raster at a 1:1 scale in the middle of the screen, which does make the display look a lot nicer, though tiny.
The new idea would be for it to center the largest integer scale of the raster that fits on the screen instead. Then your game display would look nice but be larger.
I would suspect it is trivial for the LCD to do this, probably easier than the current "blurry" scaling. And it could be a worthwhile feature some would pay for, such as you.
This is correct, applications using NeWS under the standard window manager did scale. You could also rotate windows and I would think you could do skew transformations, though I did not try it.
There was no antialiasing so sometimes the results were pretty ugly, and I also got in a big fight with somebody at Sun there because I wanted the scaling to treat the corners of pixels as the identity rather than the centers of pixels (this would allow rectangles that touch each other to remain touching as they scaled even in the round-to-nearest-pixel drawing that NeWS did), TnT got that wrong, though underlying NeWS worked great.
It is pretty sad that NeWS features over 20 years old is being reintroduced by Windows/OSX/Gnome/KDE and they are all acting like they just invented it. None of them rotate windows (other than compositing extensions but that is not exactly the same thing as the source bitmap is not a rotated image).
Generally the one-click api does not do anything until the mouse is released. Therefore it can distinguish between a "drag" and a "click". This is how they make it possible to move icons around. I'm not sure I understand your question about list boxes, it is true that lists inside a window with some kind of "OK" button do not press the OK when you click on them, but this has been true for ages. All systems I have seen with popup lists act like clicking a list item selects and dismisses the list.
I have no real opinion on this, but the alternative is that you press the right mouse button to get the "list of things to do to this icon" and then pick one, while clicking with the left mouse button would mean "do the most common operation to this icon". Therefore the alternative one-click idea is completely consistent and usable. It is however different.
I agree that "select this icon" is common enough, and that all systems no matter what they do on a click still seem to support it (on one-click systems you have to drag out a box surrounding the icon). The biggest problem with one-click is that there is no way to get into the "selected" state except by using drag.
The biggest problem with double-click is that the majority of things on the screen nowadays do an action with only a single click (ie all the buttons and links in web browsers) so it is inconsistent.
Slashdot Mirror
That's hardly comparable. Whether overpriced or not, the excuse for expensive drugs is that it pays for the development. I can assure you that the development of an ink cartridge cost a tiny fraction of the manufacturing price.
Now maybe if big Pharma built free hospitals for everybody and then relied on the prices of the drugs used by those hospitals it would be similar to what the printer companies are doing.
A 0.1 degree change from 20C is 1/2 of 1%
Holy crap! You could at least use Kelvin. Why then you could say the change is only .03 of 1% and thus really really tiny!
Instead you have proved your total ignorance of science. Thank you.
Travel to London is not too bad.
You can take the Heathrow Express, which takes about 15 minutes and costs a lot (15 pounds?). It drops you at Paddington station, which is the northwest corner of the interesting area and all the tube lines go there.
If you have more time a cheaper option that really works is to take the Picadilly line in. This will cost less than 4 pounds (get an Oyster card even if you are there for only one day), and you are already on the tube so you can get to whatever station you really need for no extra cost. However travel time is like 50 minutes. Most of it is above ground so you can see something.
You can still get plenty of news for free on the Internet!
Did you know that George Bush parachuted out of the airplanes just before they hit the WTC? I would not know that except for reading the free news. Also did you know that Al Gore is using global warming as a smokescreen to hide the thermal exhaust from his secret base under the ice cap from which he will enslave the world?
As somebody else pointed out, it is BSD for Google's own code, and GPL2/3 or LGPL2/3 for code based on projects using those licenses.
It is not using X11 in any form, it is using it's own graphics + windowing api.
My guess is that an application running on it is more like a web server, giving pages to the "browser" to render.
By hard disks they meant "storage with a moving part", not mass-storage. Flash and other solid state memory is allowed, especially removable ones. I really don't see how they could prevent external USB hard disks however, without also disabling most USB flash disks.
I would say that is a LOT of people.
My GF uses aol mail to read all her email. On several machines (OS/X) I have correctly set up Mac mail and even Thunderbird to read/write her AOL email, which worked great, and she still uses the web page to do it.
This really convinces me that there will be zero problems with putting everything into the browser. AOL mail is far worse than gmail.
According the the etymology I read the BDSM practice is named after the character in Pulp Fiction, who was named after the common use of the word "gimp".
In any case I think it is silly to say this when the primary complaint is that for a lot of people "gimp" means "cripple" or "disabled person" or whatever the term is. Bringing up a BSDM term that may very well be more obscure than "GNU Image ManiPulator" does not advance the argument.
Can other programs (such as the image viewer such as EOG that you get if you double-click an image) print reasonably correctly? If so then you have seen exactly a problem with color management: when it screws up it screws up good.
If other programs also seem to mess up in a similar way I would say the printer driver is screwed.
The question I have is why everybody complains about this when it is THE EXACT SAME THING in Photoshop? What the hell, is the fact that it is Photoshop somehow make it ok?
You seem to be posting this a lot. Actually "gimp" is slang for a disabled person (I think it may imply physical rather than mental disability). The "sexual perversion" is from a name of a bondage slave character in the movie Plup Fiction, the name obviously being chosen from the common meaning of "gimp", not the term for the bondage itself.
If you are going to question the name of GIMP then at least get your facts right.
You hold down shift and click to draw a straight line.
Please explain how you draw a straight line in Photoshop. I searched all over, because I have heard this complaint, but the only thing I found was "hold down shift and click". It is EXACTLY THE SAME (except Gimp has a overlaid line preview which makes it a lot better). I may be an idiot, but I tried all the polygonal tools and everything, nothing drew a straight line.
There is a lot of crap being posted here.
I am certainly not a professional user in any way, but have used Gimp (the one with Ubuntu 9.04) and Photoshop (most recent on OS/X) about equally.
The windows seem to be pretty much the same. Gimp creates a single one on the left, I have managed to get more than one but hate it, and it was frustrating to mess with it and try to figure out how to get it back into a single stacked window. I think they fixed things considerably by moving the menu bar from the tool window to the "main" window. Photoshop seems to create two big floating windows, one on the left and the other on the right.
I like Gimp's panel that is the current tool options. Photoshop seems to hide the options for the tools in different places depending on the tools. I really dislike the Photoshop slide-out windows on those things on the right and wish they would just put them in tabs in the floating windows like some other things.
I like the fact that you can differentiate the mouse and pen in Gimp and get three tools. Photoshop seems to think the mouse and pen tip are the same tool. And in Gimp if I use the eraser end to pick brushes, I change the eraser's brush, in Photoshop it changes the pen's brush (the only way to change the eraser seems to be to pick eraser with the pen tip, change it, then put the pen tip back). For that I give Gimp a big win in usability. I also think Gimp is enormously better at zooming with the mouse wheel, Photoshop has some very squirrly results when you do this that I really cannot explain on a modern program...
Photoshop seems to make better smoother brush strokes. Gimp had a horrible problem that antialiasing disappeared if you painted when zoomed out more than 1:1 but they seemed to have fixed that, however that sort of bug is pretty close to inexcusable, sorry. Also all brushes other than the default solid ones are useless hacks in Gimp, but appear to produce real painterly effects in Photoshop.
A huge obvious technical benefit of Photoshop is layer grouping (really a tree of compositing operations rather than just a line).
Both programs suffer from an enormous number of mystery components. What do the "chains" mean in the layers? I have clicked them on/off in both and can't see what they do. Selecting regions, cutting and pasting, and somehow getting the pasted thing to "stick" is a complete mystery.
That's a silly complaint. If you typed some text, it would have changed to the selected font as you picked it from the menu. Photoshop has an option to set the menu this way as well. It is useful because sometimes the font names are unreadable when written in the font itself, and I certainly have set it this way.
(the fact that Linux lists a whole lot of useless fonts such as a zillion ones that are really just some foreign subset of letters and don't change the main font is a bug. However I was shocked when I tried a recent version of Windows and saw identical enormous lists of useless fonts, I guess they think that was some killer feature of Linux that they wanted to copy???)
I'm reasonably certain even the cheapest photo editor that claims "red eye removal" does it by manipulating the red channel differently than the other channels. Most common solution is to set r = min(r,max(g,b)) in the locations you paint. Slightly more complex is to multiply the result so that the overall brightness remains the same, you don't even have to weight the channels differently. This is rather harmless even if you apply it to the entire image, so simply doing this to an eye-sized blob of the picture, even with no antialiasing of the edges, will work.
In any case I think even the goofballs who write the crap like the "free photo software included with your camera!" can do this correctly and would be suprised if they got it wrong.
Though dweebs here like to throw out buzzwords like CMYK and > 8 bits, the most obvious missing thing is that you cannot "group" the layers so that the compositing operation is done between them and then the result is overlayed. For instance you cannot non-destructively colorize a lineart layer and put it on top of a background, something that Photoshop makes easy.
More than 3 channels (CMYK is one minor use of that) would be nice. In professional special effects graphics these are used mostly for mattes and effects channels and information such as the normals of the surfaces. Use for the printing "black" is a minor insignificant detail compared to these other things.
Having worked with professional graphics quite a bit I have to say that "color management" is 95% bullshit. It is not possible to make a reflective printout the "same" as a light-emitting screen, anybody claiming this is lying.
Photo manipulation and painting is helped considerably by not losing information on display, this means that on current 8-bit images and 8-bit displays, any method other than 1:1 mapping of the image values to the display is WRONG, and thus most "color management" is in fact harmful (dithering and error diffusion can resolve this problem some, but nobody is doing it because users don't like the slightly-visible patterns, 10-bit displays may help here).
If you really want to manage actual light data, the most important step is to change the internal representation to a "linear" format where the emitted energy is proportional to the stored number, but the "color management" people refuse to do it because it would make "color management" (ie changing the primaries) into a trivial matrix transform and put them out of business. Also it is not practical in any integer-based storage format.
I very much hope they forget completely about any integers > 8 bits. If you are going to use 16 bits then use ILM/Nvidia "half" floating-point format. Stop living in the previous century and pretending something Photoshop did then is actually modern...
What's the difference between putting a picture into your "vacation 2009" and tagging it with "vacation" and "2009"?
The difference is that if you then decide to tag half of them with "summer" then that half are no longer in the "vacation 2009" directory.
Symbolic links can make this somewhat work, but as soon as you have a number of tags that form intersecting sets the number of directory entries grows to m*2^n (where m is the number of files and n is the number of different tags). This is far larger than the m*n space needed by tags even with a stupid implementation.
I do however believe a working system can be built where POSIX file api still can be used. A tag is a directory, and a whole tree of them is just a directory listing the files in that. So ./vacation/2009/ and ./2009/vacation/ would both have the same files: all the ones tagged with those. ./vacation/2009/summer would have files tagged with all three.
This is exactly what should be done. But not with "metadata" as people are saying. We need "metadata" embedded directly into the files. Then it is preserved when the file is copied, archieve, or stored on a system that has a different concept of metadata.
The normal implementation of "metadata" would then be used to cache the metadata read out of the file. For the best example of how this works, imagine the "thumbnail images" that some systems are storing. These are stored in the file system in some way, but are completely figured out from the original file. If the thumbnail metadata disappeared it would get repliated again later. What I would like to see is all metadata is produced from the file contents in this same way.
The way to do it is to make up some rules for how to find it, based on the file types. This does require a program that understands a lot of file formats, but that would be an actual executable program, not a "library" and thus easy to expand and for any program to call. It would for instance understand how to extract jpeg comments. For text files it would know how to recognize comments and then rules could be made up for patterns in the comments to indicate the metadata.
I could see this eventually replacing all "metadata" including long-standing things such as the name, modification date, and even the protection bits of the file (the actual protection would be the intersection of some guarded protection and what the file says, but a program with sufficient privledges could update the real protection to match the file data).
When offering something up like this is the company expected to just put up a big banner at the top saying, "HEY, WE ARE CHARGING YOU FOR SOMETHING IF YOU CLICK YES!" before even trying to sell the person on the product?
YES! That is an excellent description of exactly what should be required.
However the upscaling is usually used to show the Windows startup screen, so an improved upscaler (or just making it center a 1:1 image in the middle of the screen) would make a lot of buyers go "hey this screen looks a lot better" when they turn their computer on.
So I agree it is a mystery why they go for such fuzzy awful upscaling.
A lot of LCD screens have an option to center the raster at a 1:1 scale in the middle of the screen, which does make the display look a lot nicer, though tiny.
The new idea would be for it to center the largest integer scale of the raster that fits on the screen instead. Then your game display would look nice but be larger.
I would suspect it is trivial for the LCD to do this, probably easier than the current "blurry" scaling. And it could be a worthwhile feature some would pay for, such as you.
This is correct, applications using NeWS under the standard window manager did scale. You could also rotate windows and I would think you could do skew transformations, though I did not try it.
There was no antialiasing so sometimes the results were pretty ugly, and I also got in a big fight with somebody at Sun there because I wanted the scaling to treat the corners of pixels as the identity rather than the centers of pixels (this would allow rectangles that touch each other to remain touching as they scaled even in the round-to-nearest-pixel drawing that NeWS did), TnT got that wrong, though underlying NeWS worked great.
It is pretty sad that NeWS features over 20 years old is being reintroduced by Windows/OSX/Gnome/KDE and they are all acting like they just invented it. None of them rotate windows (other than compositing extensions but that is not exactly the same thing as the source bitmap is not a rotated image).
Generally the one-click api does not do anything until the mouse is released. Therefore it can distinguish between a "drag" and a "click". This is how they make it possible to move icons around. I'm not sure I understand your question about list boxes, it is true that lists inside a window with some kind of "OK" button do not press the OK when you click on them, but this has been true for ages. All systems I have seen with popup lists act like clicking a list item selects and dismisses the list.
I have no real opinion on this, but the alternative is that you press the right mouse button to get the "list of things to do to this icon" and then pick one, while clicking with the left mouse button would mean "do the most common operation to this icon". Therefore the alternative one-click idea is completely consistent and usable. It is however different.
I agree that "select this icon" is common enough, and that all systems no matter what they do on a click still seem to support it (on one-click systems you have to drag out a box surrounding the icon). The biggest problem with one-click is that there is no way to get into the "selected" state except by using drag.
The biggest problem with double-click is that the majority of things on the screen nowadays do an action with only a single click (ie all the buttons and links in web browsers) so it is inconsistent.